In recent years, it has become common for a user to input characters aurally due to improvements in speech recognition technology. However, utterance information included in sounds emitted by a user often includes unknown words not registered in a database (e.g., a corpus or the like), such as abbreviations, paraphrases, and proper nouns.
Therefore, unknown words included in utterance information of the user must be properly recognized in order to convert the utterance information of the user into proper notation. Recognized unknown words are registered in a database, and so will thereafter be converted into proper notation.
An example of such technology may be technology which, in continuous dialogue between users, extracts unknown words from dialogue content and registers the unknown words in a predetermined database without making the user aware of the registration mode, as described in Patent Literature 1 below, for example.